Heavy alcohol use and its associated problems are a serious public health issue among young adults, with college students residing on campus showing the highest rates of hazardous use. The college environment in which peer relationships are highly valued exacerbates risks seen in emerging adulthood, but intervention approaches that use the peer social network to induce change in heavy drinking have not been developed for college student populations. Social network theory suggests that an empirically effective intervention administered to a minority of prominent social network members can result in change throughout the network as a function of ties in the network. Social learning theory provides a complementary perspective that articulates specific mechanisms through which the influence of behavior change in a network may occur. The objective of this proposal is to establish the efficacy of reducing heavy drinking and alcohol problems in college students by reducing the heavy drinking of influential members in their peer social network. The specific aims are to: (1) investigate the efficacy of targeted Brief Motivational Intervention for reducing heavy drinking in network members who received no intervention; (2) identify the social influence mechanisms through which the intervention effect was conveyed; (3) identify network and relationship features that moderate intervention efficacy; and (4) investigate the intervention effect on participant network position and drinking-based selection within the network. Building from successful pilot work, first-year students (N = 1,280) living in 10 first-year dormitories on one college campus will be enrolled and assessed in the first month of the year. Assessments will include behavioral measures and network ties to other students in the same class year. Dormitories will be paired by size and randomly assigned to a Brief Motivational Intervention (BMI) or Natural History Control (NHC). Using baseline data, heavy drinking participants in the top quartile of betweenness centrality - an indicator of social influence in the network - (n = 160 12.5% of participants in each intervention group) will receive a single session BMI midway through the first academic semester or will serve as controls. Follow-up assessments will be conducted on all participants 5 and 12 months post-baseline and the indirect effect of the intervention will be tested on the other heavy drinkers in the sample (N = 480) who received no direct intervention. Five domains of social influence derived from social learning theory will be investigated as mechanisms of the transmitted intervention effect, and network proximity and relationship quality will be investigated as moderators. Multilevel modeling will be used to evaluate intervention outcomes, and stochastic actor-oriented modeling will be used to examine hypothesized social influence mechanisms, moderators, and changes in the network. This research design represents a bold step in alcohol preventative intervention research and has the potential to inform the development of interventions that have effects beyond the intervention recipients themselves.